Vehicular Communications and Networks

Maziar Nekovee

BT Research & University College London

[email protected]
Outline
• Technology trends and applications scenarios

• Characteristics, challenges and architectures for IVC (high mobility broadband wireless
access on the road)
• Characteristics and challenges of vehicular adhoc networks (safety and ITS applications)
• Vehicular mobility modelling
• ITS case studies
– Information dissemination in highly dynamic, intermittently connected VANET
– Traffic congestion reduction using V2V communications
– V2V-based rear-end collision avoidance
– Parallel simulations of large V2V networks
– Cyber-attacks on VANET

• Emerging technologies: dynamic spectrum access, cognitive radios, smart and directional
antennas

• Summary

• Bibliography

2
Technology trends
• BMW, Mercedes, Fiat , Ford, Toyota, Nissan, … are prototyping vehicles equipped with
Wi-Fi (802.11a/b/g) and DSRC (802.11p)

• Wi-Fi enabled vehicles are expected to be on the road within the next 3-5 years

• In the US, FCC allocated 75 MHz of spectrum for dedicated short-range vehicular
communications (total UK 3G spectrum is ~ 140 MHz)
• In the UK and across the EU 30 MHZ of spectrum has been put aside for vehicular
networks.

• Assuming 10% market penetration, this amounts to ~3-4 million Wi-Fi enabled vehicles
carrying people in the UK, and ~15 million in the US

• Vehicles equipped with wireless technology can communicate directly with each other (V2V),
and with the fixed infrastructure (V2I). They can form Vehicular Adhoc Networks (VANET)
• New opportunities in:
– High mobility broadband wireless access
– Intelligent Transport Systems (ITS)
– Sensor networks on the road

3
High mobility broadband wireless access

• Extending broadband access

to high mobility users in cars,
buses, coaches in cities,
highways, roads Gass, Scott, Diot, Intel Research, 2003.

– Mobile office (Internet, email, file

transfers ..)
– Entertainment (video-on-demand,
games ..)
– Location based services and
charging (parking, road usage,
local restaurants, local traffic info,
dynamic map updates, ..)
– Voice over IP (VoiP)
Not exclusively WiFi but a hybrid of
3G/LTE, WiFi and mobile WiMAX
Ko, Sim, Nekovee, 2006
4
Safety and Intelligent Transport Systems

• Improve road safety, increase

efficiency of road usage, reduce
congestion and traffic jams
– Early warning of road hazards
– Driver assistance and collision
avoidance
– Real-time traffic monitoring and control
on a much finer scale than is possible
now (with loop detectors)
– Real-time route guidance and journey
planning
– Cooperative driving: lane merger, high-
speed platoons
– Real-time traffic control and re-
shaping/smoothing

5
Final Report, Traffimatics Project , 2006
 Sensor networks on the road
• Position sensors
– GPS, accelerometer, compass, tilt
sensor
• Environment sensors
– CO2, cameras, thermometer,
barometer, humidity sensor
• Vehicle sensors
– ignition, speed, engine speed, engine
temperature, … Cambridge sentient van
• Vehicle interior sensors
– camera, ID card reader
• Wireless communication
– 802.11{a,b,g}, GPRS, 3G

Source: Davies, Cottingham, Jones: A Sensor Platform for

Sentient Transportation Research, LNCS 4272. Oct. 2006.

Cellular coverage as mapped

by Cambridge sentient van 6
Timelines

T. Kosch, BMW R&D, 2005. 7

High Mobility Broadband Wireless
Access (BWA)

8
 BWA - Wireless technologies
3G DSRC Wi-Fi
802.11p ( WAVE) 802.11a/b/g
Range up to 12 km up to 1 km up to1km

data rate 384 Kbps – 2Mbps, up to 27 Mbps up to 54 Mbps

5.76/ 14.4Mbps
(HSUPA/ HSDPA)
Spectrum / GHz 1.8, 1.9 and 2.1 5.9 (USA), 2.4 (b/g), 5.2, 5.8 (a)
5.8 (Japan, Europe)
licence licensed dedicated spectrum licence-exempt*
(USA)

access mechanism centrally scheduled contention based contention based

limitations •Very high deployment •short to medium range •short to medium range
costs •Interference issues
•low transmission rates, due to shared spectrum
scalability (backhaul)
advantage •already available, •low deployment costs •low deployment cost,
•large coverage •distributed •distributed

9
 BWA - Wireless technologies
WiMAX 802.16e WiBro LTE
(Nomadic) (Mobile, <=
60km/h)
Range up to 5 Km up to 5 Km up to 12 km

data rate maximum 70 Mbps; maximum 50 Mbps 20-100 Mbps

10 Mbps @ 10km
spectrum 2.3/2.5, 3.5, 5 GHz 2.3 ~ 2.4 GHz ?? (digital dividend, 800
MHz band, e.g.
Germany)
licence licensed & licence- licensed licensed
exempt

access centrally scheduled centrally scheduled centrally scheduled

mechanism
limitations •high deployment •high deployment •high deployment costs
costs costs •Scalability (backhaul)
•scalability (backhaul) •scalability (backhaul)

advantage •high data rates •high data rates •High data rates
•large coverage •large coverage •large coverage 10
System Architecture
Infrastructure
Internet MESH

Gateway

V2V forming I2V or roadside-

VANET to-vehicle

Legend: Road site AP/BS Backhaul wireless mesh router

(omnidirectional antenna) (directional antenna)

11
WiFi vs cellular based architecture

Wi-Fi Ko, Sim, Nekovee, 2006. Cellular

12
Performance of Wi-Fi-based BWA

™Vehicle-to-vehicle communications
™ new channel model and performance study

™Single vehicle / single roadside AP

™ 802.11 performance at vehicular speeds (urban and highway)

™Single vehicle / multiple roadside APs

™ Co-channel interference and optimal channel allocation
™ Handoff at speeds

™ Many vehicles /multiple roadside APs

™ (Rapid) temporal (day/night/rush hour) and spatio-temporal (traffic
jams) change in user density Æ Requires highly adaptive protocols

13
 Vehicle-to-vehicle communications
• Measured power spectrum of V2V • In the I2V communication channel
channel showed significant differences model: there are only scatterers near
from that of the conventional I2V the vehicle
channel • In V2V communications, there are
– The Jake’s fading channel model will not scatterers around both the transmitter
be suitable and receiver.
• Correlated double ring scattering
channel mathematical model has been
proposed for V2V
• Sum of sinusoids approximation
simulation model for V2V
Source: Acosta, G , et al, IEEE Globecom 2004

Measurement result
showed that there is
significant
performance
degradation when 2 Source : LC Wang, YH Cheng, IEEEVTC 2005 Spring

vehicles move at high

relative velocity More performance analysis is required.
Source: Singh, J.P, et al, IEEE VTC 2002 Fall
14
 Single vehicle/single AP (highway)
R x0

xo + R
1 xo + R
Total throughput: Ttotal = ∫ v
xo − R Ti ( x(t ), v )dt ≈ ∫ Ti ( x )dx
v x o − R
v

1 xo + R
Ti ( x )dx
2R
Ttotal ≈
{v
Ti , Ti = ∫
2R ox − R

residence time
15
802.11b at speeds I

6 180
Throughput
Link Quality
Keystroke Plot 160
5
140
Throughput (Mbits/sec)

4 120

Link Quality
100
3
80

2 60

40
1
20

0 0
-1500 -1000 -500 0 500 1000 1500
Displacement from AP (m)

BT Technical Report, 2005. 16

802.11b at speeds II: connectivity phases

• Experiments performed in
highway conditions
• Roof-mounted external
antenna
• UDP and TCP measurements
for both V2I and I2V
scenarios
• Bell-shaped throughput
curves (entry, production, exit
phases)
• Velocity-dependence is
mainly due to the total
residence time
J. Ott, D. Kutscher, 2004. 17
802.11b at speeds II: speed dependence

• Experiments performed under

no-interference conditions
(desert)
• External antenna on the
roof
• UDP, TCP, HTTP
• Observed some velocity-
dependent packet loss

Gass, Scott, Dio, 2003.

18
802.11a at low speeds: Suburban
• Drive past an access
point
• Highly variable
performance
• Low speeds
impacted by time
spent in nulls of radio
coverage
• Impacts data rates
for slow moving
traffic

Source: Cottingham, Wassell, Harle, VTC 2007.

19
Single vehicle/ multiple access points I

Co-channel interference:
ƒ Optimal channel assignment
is required to minimize
interference while maximizing
coverage
2 non-overlapping channels
ƒ For 802.11.b/g there are only
3 non-overlapping channels
ƒ Relatively simple in 1D
geometries (i.e. stretch of
highway)
ƒ Hard problem in 2D (graph
colouring) Æ heuristics
3 non-overlapping channels
20
 Single vehicle/ multiple access points II
802.11 Handoff at speeds
• Autonomous handoff initiated by client itself
(unlike the centralised approach in cellular
systems):
- SNR/packet loss monitoring
- Channel scanning for new AP
- Authentication/Association
• Handoff can take up to 500 ms, mainly due to
the scanning phase
• Main effect is on real-time applications (VoiP)
• Fast layer 2 handover

Source: Ramani and Savage, 2005. 21

Components of L2 handoff time

Layer 2 handoff components: • Major phases of

handover process:
HANDOFF – Detection
TIME
PHASE
– Searching
Detection* About 1000 ms – Execution
Search/
350-500 ms
Scanning
Authentication
<20 ms
& Association
Infrastructure
<20 ms
update

AP beacon frequency ~ 100 ms (default)

* Some implementations skip the detection
phase

22
 L2 handoff time improvement
• Detection
– Reduce beacon interval to 60ms Improvement can be achieved by
modifying wireless card driver!
– Reduce no. of retransmission trials

• Search
– Scan only subset of channels
– Fast scanning (reduce beacon
frequency)
– Continuous scanning (before the
need aroused)
– In VCNs road topology information
and mobility prediction can be
utilized to reduce scanning times
Source: http://user.informatik.uni-
Source: Héctor Velayos and Gunnar Karlsson ( goettingen.de/~mobiarch/2006/MobiArch0
http://web.it.kth.se/~hvelayos/papers/TRITA-IMIT- 6_Keynote_Henning.pdf
LCN%20R%2003-02%20Handover%20in%20IEEE%20802.pdf)

23
Many vehicles/ multiple access points

free flow

congestion

• Average throughput per user

depends strongly on traffic
conditions

• Requires adaptive resource

management on a much finer
time and length scale than 2G/3G

24
 Smart / directional antenna
• In the MobiSteer project, Vishnu
Navda, et. al. show that beam
steering can improve the
connectivity duration as well as
PHY-layer data rate due to
better SNR provisioning.
• Vishnu Navda, et. al., ACM • Directional antenna
MobiSys’07, June 11~14, 2007,
San Juan, Puerto Rico, USA. – reducing the transmission collisions, as
well as increasing the channel reuse
possibility.
– Suitable for VANET with parallel
neighboring vehicular traffic

25
Characteristics and Challenges of
Vehicular Adhoc Networks
(Safety and ITS Applications)

26
 Rapidly changing network topologies
• Vehicles continuously move in
and out of each other’s range
ƒ short link lifetimes
ƒ No continuous end-to-end
connectivity
ƒ Frequent network
fragmentations into isolated
clusters

2r
tlink ~ same direction Results are for average traffic in a highway (CA, US),
|Vi − V j | simulations studies by Blum and Eskandarian, 2004.
2r
tlink ~ opposite direction
(Vi + V j )

27
 Large node density variations
network fragments
• Node density variations are into isolated clusters
governed by traffic conditions: Vcritical
ƒ day/night/rush hour variations
ƒ Free flow vs. congestion
ƒ Traffic jam waves (time and
space)

congestion

free flow

end-to-end connectivity as function of mean

velocity, Nekovee, VTC 2006

28
Inter-vehicle communications at speeds

• Unlike I2V only very few (published)

measurements.
ƒ Singh et al discuss experimental test for
802.11b with two vehicles driving in
urban, suburban and highway (roof top
external antennas)

ƒ Yin et al discuss simulation studies using

a detailed radio model of DSRC , finding
some speed-dependence in the relation
between SNR and BER

ƒ Speed-dependence especially important

in opposite-lane communication scenarios

Sim, Nekovee, Ko, 2005 29

MAC layer and scalability

• To avoid interference caused by nearby

devices using the same channel, access to
medium is regulated by the 802.11 MAC
protocol.
• 802.11 uses a contention-based access
mechanism.
• Devices refrain from transmission and number of interfered neighbours ~ 2ri ρ ( x, t )
backoff for a random time when they
sense a busy medium. This can greatly
limit network throughput:

Tmax ~ Br L / 2ri
• Potential solutions
zero MAC
ƒMore channels, more MAC
spectrum (cognitive radio).
ƒTDMA-based MAC protocols
and scheduling algorithms Multi-hop broadcast in VANET, Nekovee et al, 30
(require synchronization). Proc NAEC 2005.
Vehicular adhoc communications for ITS

™Point-to-point routing

™Single-hop communications
™ Collision avoidance
™ Platoon formation

™Multihop communications
™ Information dissemination (traffic conditions, road hazards)
™ Traffic data collection

31
Point-to-point routing
• Rapidly changing topology:
– Proactive routing problematic due
to many invalid routes
– Reactive routing attempt to
discover routes when needed but
route discovery could take longer
than route lifetime!
– Location-based routing delivers
only to a zone of relevance
(shorter hops) but requires
location information (Briesmeister
et al, 2000)
– Alternatives: Infrastructure-
assisted routing and Blum, Eskandarian, Hoffman, 2004.
epidemic/opportunistic routing
(effectively point-to-multipoint)

32
 Single hop communications I
• Fog conditions. Car 2 is
driving behind car 1. ( x2 , v2 )
• Car 1 suddenly breaks and
broadcasts a warning
( x1 , v1 )
message.
• To avoid a collision we must
have:

| x2 − x1 | v2
tcomm +treaction < −c
v2 f
driver’s reaction time (~750 ms) road’s grip coefficient
message communication latency
Masini, Fontana, Verdone, 2006 (for cellular).
33
 Single-hop communications II

• Simulation studies of single-hope

communication for delay-critical safety
applications by Yin et al (2004).
• DSRC standard (PHY/ MAC protocol stack)
at 5.9 GHz is used.
• VANET created by 100 vehicles moving on
a city roadmap, each having ~1000 m
transmission range, using a single 10 MHz
common channel.
• 1-10 safety message/second/vehicle (100-
200 B) .

ƒ Performance metrics used:

ƒ Single-hop latency
ƒ Single-hop throughput (fraction of single-hop
neighbours receiving the message).

A 100 ms latency limit is proposed by Vehicular Safety

Communications Consortium (VSCC)

34
Modelling vehicular movements

• Vehicular movements in traffic are highly

correlated and are constrained by road
topology and traffic lights etc
• Conventional MANET mobility models,
random walk and random waypoint are
highly inadequate
• Fortunately there is much research
(physicists, transport engineers) on traffic
and vehicular mobility models
• Broadly speaking, there are three levels of
modelling:
• Fluid models (macro scale)
• Cellular automata models (meso
scale)
• Car-following models (micro scale)
Traffic jam waves in a highway corridor from car-following
simulations, Vazquez-Prada and Nekovee, 2005.

35
• Fluid models
• Vehicular traffic is described as an incompressible fluid (view from
an airplane)
• Characterised by: ρ ( x, t ), V ( x, t ), Q( x, t )
• Navier-Stokes-type equations describe time-evolution
• In steady-state we have (highway):
vehicle length

β
⎛ ρe ( x) ⎞ 1
Ve ( x) ≈ V f ⎜1 − ⎟⎟ d =l+
⎜
⎝ ρ jam ⎠ ρe ( x)

free flow speed Inter-vehicle distance

36
• Car-following models
Fi (t )
• Track the movements of individual
cars in time and space. ( xi (t ), vi (t ))
• Cars accelerate/decelerate due to a
force which depends on their
distance and velocity relative to the ( xi +1 (t ), vi +1 (t ))
car ahead of them.
• Additional rules for multiple-lane
scenarios.
• Each car can have its own ⎡ ⎛ Δv (t ) ⎞ 4 ⎛ x 0 ⎞ 2 ⎤
individual attributes (length, driver Fi (t ) ~ ⎢1 − ⎜ i0 ⎟ − ⎜ i ⎟ ⎥
behaviour, car/truck …) ⎢⎣ ⎝ vi ⎠ ⎝ Δxi (t ) ⎠ ⎥⎦
• Can accurately reproduce real
traffic behaviour in highways and
urban scenarios. vi0 : desired speed (speed limit)
Δvi (t ) = vi +1 (t ) − vi (t )
Δxi (t ) = xi +1 (t ) − xi (t ) − l

37
 Coupled simulation approach
(Large scale experimental evaluation is not an option)

vehicles’ movement & road topology

microscopic vehicular traffic simulator (IDM,
control data packers
Traffic information and

Dracula, TranSim)

wireless network simulator (Trafficom, NS2,

NS3)

Grid Computing Platforms (Legend, Hector, NGS)

38
Information dissemination in VANET I

39
Information dissemination in VANET II

• Applications:
– Local traffic conditions for ITS.
– Warning messages (road hazards, accidents, congestion)
– Sensor data alerts.
– Epidemic routing.
• Challenges
– Intermittent network connectivity (reliability issues).
– Excessive network traffic and MAC latency caused by highly correlated
transmissions (scalability issues).
• Proposed approaches.
– Infrastructure-assisted: roadside info-stations/accesspoints/cellular assist
VANET to bridge the gaps.
– Purely ad-hoc: store and forward/opportunistic mechanisms similar to those
used delay-tolerant networks.
– Selective broadcasting schemes (deterministic, probabilistic).

40
Information dissemination in VANET III

• Limitations
– Infrastructure-assisted: Infrastructure may not be always available,
single point of failure
– Adhoc: often requires control data exchange (e.g. to maintain
clusters), or additional information (e.g. road topology information
and location)
– Selective broadcasting schemes address scalability but cannot cope
with intermittent connectivity and network fragmentations

41
Edge-aware epidemic
• Persistent flooding achieves
100% reliability but generates Nekovee and Bogason, 2007.
excessive traffic.
• In epidemic protocols nodes re-
transmit messages with a
probability P.
cluster edge
• This reduces traffic but reliability
is probabilistic (even in static
networks).
• Edge-aware epidemic:
Only nodes at the edge of a
cluster keep the message alive.
message source
• How does a node know it is on
the edge?

Source: Nekovee, IET Intelligent Transport , 2009 42

 Algorithms
• After receiving a new message a node
selects a backoff time from [0,Tmax] and
waits.

• When the waiting time expires it counts the

received messages from vehicles in the
front, Nf, and from the back, Nb.

• It then makes a probabilistic forwarding

decision based on the imbalance.

• Only nodes at the edge “survive”.

• They periodically broadcast the message

until there is a cluster merger.

• Directional messaging can be handled in

the same way (cluster head and tails).

43
IDM -VANET simulator

http://nekovee.info/VANET1.htm
http://vwisb7.vkw.tu-dresden.de/~treiber/MicroApplet/
44
 Scalability
• Road with one lane in each
direction
• High vehicle density/lane Æ
continuous e2e connectivity
• Transmission range:120 m
• We inject a message in a
randomly chosen vehicle and
follow its propagation
• Results averaged over a large
number of simulation runs

45
Reliability
• Road with one lane in each
direction
• Vehicles move at a specific flow
into the road
• Flow rates was adjusted to obtain
intermittently connected networks
• Message injected at t=30 s
• Transmission ranges : 60, 120 m
• Both omni-directional and
directional propagation scenarios

46
 Congestion reduction using V2V
messaging

Accident

Source: Hewer and Nekovee, EURASIP Journal Adv. Signal Processing, 2010

47
100% WiFi-equipped vehicles

48
49
V2V communication for rear-end collision
avoidance

• Better that conventional break light (visibility, LOSÆ chain collisions)

• Better than cellular (distributedÆ faster and cheaper).
cellular delivery latency ~ 1-4 seconds
• Suffers from communication delay (802.11p’s MAC contention
mechanism)
• Suffers from packet loss (hidden node + MAC contention+ wireless V2V
channel)
.

50
Performance modelling of V2V-based rear-
end collision avoidance protocols
• V1 is moving ahead of V2.
• V1 suddenly bakes to avoid a hazard.
• Upon braking a warning message is
triggered, and is broadcasted using V2V
adhoc communication ( xi (t ), vi (t ))
• In principle superior to brake lights
signalling (low visibility, driver’s slow
reaction).
• Reduces the chance of chain collision due
to increased “visibility” range. ( xi +1 (t ), vi +1 (t ))
• In practice V2V communication is subject
to delivery latency and packet lossÆ
repeated retransmission

• A precise formulation of QoS requirements

for collision avoidance not available in
literature
• We provide analytical results to guide
V2V protocol design.

51
 Maximum delivery latency and minimum retransmit
frequency driver reaction time
Inter-vehicle gap

emergency deceleration

Single hop delivery latency Single hop packet loss rate

52
 Modelling ingredients
• Realistic car-following model (IDM) for vehicular movement:
Random walk and random waypoint mobility models cannot describe the highly correlated motion
of cars (Treiber et al, Phys. Rev. E, 2000).
average velocity per lane

average distance
−1/ 2
⎡ ⎛ V ⎞δ ⎤
between adjacent cars

S (V ) = ( S0 + VT ) ⎢1 − ⎜ ⎟ ⎥
⎢⎣ ⎝ V0 ⎠ ⎥⎦
average density per lane
1
ρ=
S +l
• Velocity-dependent wireless channel model:
In highways relative velocity difference between two communicating cars can reach 200-300 Km/h
(in opposite lanes).
.
signal attenuation: distance + velocity+ log normal shadow
signal attenuation threshold fading (Sim, Nekovee, Ko, 2005)

1 1 ⎛ β th − β%(rij , vij ) ⎞
pij = pij (r ij , vij ) = + erf ⎜ ⎟
2 2 ⎝ 2σ ⎠ 53
Maximum acceptable latency (no loss)

free flow
traffic jam 54
Minimum message retransmission frequency

without shadowing & fading with shadowing & fading

55
Next steps - M25 London
• M25 London Orbital

• 121.5 miles long

• Longest and one of the most

congested ring roads in the
world

• 31 junctions

• 9 motorway interchanges

• Junction 15 to 14 carries 165000

cars per day

• Simulating just Junction 15 to

14 for 24 hours would take
over a year to achieve on a
single processor machine

56
 Parallel simulations of V2V networks
•Run many (10,000) independent
simulations of small networks on parallel
machines with different parameter combinations:
• vehicle density
•packet size,
•transmit rate,
•transmit power
• The aim is to map out V2V network
performance metrics (latency, packet drop rate
etc) as multi-variable functions of system
parameters.
•Design traffic- adaptive V2V protocols

57
 Cyber-attacks on VANET
Securing vehicular networks against cyber-attacks is a major
issue due to the criticality of many of their applications in ITS.

58
 A plethora of potential cyber-attacks

Raya and Hubaux, 2005.

Wireless worm attacks

Blum and Eskandarian, 2004.
Nekovee, 2006.
Khayam and Rada, 2004.
59
Modelling worm attacks on VANET

• Worms are self-replicating

computer viruses that can
spread in networks without any
March 2005
human intervention.

• The last few years has seen the

emergence of new types of
worms that specifically target
wireless networks. .

• Unlike conventional worms, they

can propagate wirelessly, using a
radio communication technology,
such as WiFi or Bluetooth.

Nekovee, 2006
60
Worm spreading model
• A worm spreads in the network by multi-hop forwarding, instead
of using IP scanning which could be highly inefficient for
VANET.
• Each car can be in one of three states:
susceptible, infected, patched.
• Infected cars broadcast the worm at any possible opportunity.
• Susceptible cars become infected at a rate lambda when
they receive a transmission from an infected car.
• Infected cars can get patched at a rate delta.
• The time it takes to process a worm by each car is a constant
value of one clock tick. Worm transmission is assumed to be
instantaneous.

61
Simulation studies
• We consider worm propagation in a 10 km highway corridor
consisting of 1 lane in each direction.

• Effective transmission range per vehicle is fixed at 250m.

• Traffic conditions are characterized by mean velocity in each

lane, e.g. V=0,20,40.60,80,100,120 Km/h and the IDM model is
used to compute the mean vehicle density.

• For a given density we create an instant of the vehicular

network by distributing vehicles randomly and uniformly in each
lane, and linking two vehicles (i and j) with a probability given
by our velocity-dependent shadow fading model.

• A randomly chosen vehicle is infected by the worm at the

beginning of a each simulation .
62
Propagation dynamics of an unknown worm in VANET

network fragments

Standard Worm
Propagation Model

63
Impact of interactive patching

64
Emerging trends/ technologies for
spectrum access
ƒ Radio Spectrum is a key resource for
wireless networks
ƒ In the past spectrum was rigidly
regulated and managed
The FCC Frequency Allocation Table at 3GHz.
ƒ But this is changing to a much more
dynamic management of and access
to spectrum (phenomenal increase in
spectrum+ inefficiency of command &
control + technology advances)
ƒ Emerging technologies:
ƒ Software-defined radio (SDR)
ƒ Dynamic Spectrum Access (DSA)
ƒ Cognitive radio (CR)

A mobile phone conversation using a cognitive radio

link
65
Scientific American, March 2006.
Dynamic Spectrum Access

• Dynamic Spectrum Access (DSA) techniques allow devices (just-in-time)

selection of the most appropriate spectrum
• Early forms of DSA already exist
• BT Fusion phone is an example:
ƒ Indoor: Wi-Fi (licence-exempt)
ƒ Outdoor Cellular (licensed)

66
Cognitive radio

• Smart devices built on

software-defined radio: licensed band Spectrum gaps

– scan their environment.

– identify unused (licensed)
spectrum bands (learning).
– move into one of these bands
(frequency-agile/reasoning).
– move out to another band
when necessary (predictive).

The Trajectory of a cognitive radio in

time-frequency space, Nekovee 2006.

67
Summary
• VCNs hold promises for a plethora of important applications:
– High Mobility Broadband Wireless Access
– Future Intelligent Transport Systems
– Pervasive Sensor Networks on the Road
• A tough but exciting area of research at the intersection of a number of
disciplines and technologies
• Important advances have been made in research but many open research
challenges:
– Handoff at speeds
– Traffic-adaptive protocols
– Scalability
– Security
– Spectrum demand and interference management
• Advanced simulations and modelling coupled to measurements are essential in
order to address research challenges in the realistic context of large scale
systems
• They can also to give us a glimpse of the future

68
69
Further readings

‰ Blum J, Eskandarian A. Hoffmann L, Challenges of intervehicle adhoc networks, IEEE Trans. Intelligent
Transportation Systems, Vol. 5, No. 4, 2004, pp 374-351.
‰ Blum J. and Eskandarian A, The Threat of intelligent collisions, IT Pro, January/February 2004, pp 24-
28.
‰ Cottingham D N, Wassell I, Harle K R, Performance of IEEE 802.11a in vehicular contexts, Proc. IEEE
VTC 2007( Spring), 2007.
‰ Davies J J , Cottingham D N, Jones B D, A: A Sensor Platform for Sentient Transportation Research,
LNCS 4272. Oct. 2006.
‰ Frusler H. et al., A comparison of routing strategies in vehicular adhoc networks, 2002.
‰ Gass R, Scott J, and Diot C, Measurements of In-motion 802.11 networking, Intel Research Report, IRC-
TR-05-050, 2003.
‰ Kosch T, Technical concepts and prerequisites of car-to-car communications, 5th European Congress
and Exhibition on ITS, 2005
‰ Ko Y F, Sim M L and Nekovee M, Wi-Fi based broadband wireless access for users on the road, BT
Technology Journal, Vol. 24, No. 2, 2006.
‰ Little T. D. C and Agarwal, A, An information propagation scheme for VANETs, Proc. 8th IEEE Conf. on
Intelligent Transportation Systems, Vienna, Austria, 2005.
‰ Masini M, Fontana C and Verdone R, Provision of an emergency warning service through GPRS:
Performance evaluation, Proc. IEEE Intelligent Transportation Systems Conference, 2004.
‰ Nathan, B. and G. Jinhua, Increasing broadcast reliability in vehicular ad hoc networks, in Proceedings of
the 3rd international workshop on Vehicular ad hoc networks. 2006, ACM Press: Los Angeles, CA, USA.
‰ Nekovee M, Sensor Networks on the road: The promises and challenges of vehicular wireless networks
and vehicular grids, Proc. 1st Workshop on Pervasive Computing and e-Research, 2005.
‰ Nekovee M, Modelling the spread of worm epidemics in vehicular adhoc networks, Proc. IEEE VTC06
(Spring), 2006.
70
‰ Nekovee, Dynamic spectrum access -- concepts and future architectures, BT Technology Journal,
Volume 24, no. 2, 2006, pp 111-116.
‰ Nekovee M and Bogason B, Reliable and efficient information dissemination in intermittently connected
vehicular adhoc networks, Proc. IEEE VTC07 (Spring), 2007.
‰ Ott, J. , Kutscher D, Drive-thru Internet: IEEE 802.11b for “Automobile” users, Proc. IEEE INFOCOM,
2004.
‰ Ramani I and Savage S, SyncScan: Practical Handoff for 802.11 Infrastructure Networks, Proc. IEEE
INFOCOM, 2005.
‰ Raya M and Hubaux J-P, Securing Vehicular Ad Hoc Networks, Journal of Computer Security, Special
Issue on Security of Ad Hoc and Sensor Networks, Vol. 15, Nr. 1, 2007, pp. 39 - 68.
‰ Sim M L, Nekovee M and Ko Y F, Throughput analysis of Wi-Fi based broadband access for mobile
users on the highway, Proc. IEEE International Conference on networks, 2005.
‰ Ko Y F, Sim M L, Nekovee M, IEEE 802.11b based broadband wireless access for users on the road, BT
Technology Journal, Vol. 24 No2, 2006, pp 123-129.
‰ Singh, J P, et al. Wireless LAN performance under varied stress conditions in vehicular traffic scenarios.
in Vehicular Technology Conference, 2002. Proceedings. VTC02 (Fall), 2002.
‰ Schroth C. et al. Simulating the traffic effects of vehick-to-vehcile messaging systems,5th International
Conference on ITS Telecommunications, 2005.
‰ Treiber M, Hennecke A and Helbig D, Congested traffic states in empirical observations and microscopic
simulations, Physical Review E, Vol 62, 2000, pp 1805-1823.
‰ Wu H, Fujimoto R, Guensler R, Hunter M. MDDV: A Mobility-Centric Data Dissemination Algorithm for
Vehicular Networks, Proc. 1st ACM international workshop on Vehicular ad hoc networks, 2004.
‰ Yin J et al. Performance evaluation of safety applications in DSRC vehicular adhoc networks, Proc. IEEE
VANET’04, 2004.

71
• Thanks and acknowledgements
– David Cottingham, The University of Cambridge
– Stephen Eubank, Georgia Tech
– Keith Briggs, BT Research

– Video: Courtesy of Synthetic Data Products for Societal Infrastructures and Proto-Populations: Data
Set 3.0, NDSSL-TR-07-010, Network Dynamics and Simulation Science Laboratory, Virginia
Polytechnic Institute and State University, 1880 Pratt Dr, Building XV, Blacksburg, VA, 24061,
ndssl.vbi.vt.edu/Publications/ndssl-tr-07-010.pdf

72
 Performance of VLC receivers
0
10 0
10
-5
• Vary over the day
1010-5
– Due to variation in
-10
10 -10
10
background light/
-15
1010-15
global irradiance
-20
1010-20
Performance
• Newly proposed
BER

Threshold
-25
1010-25
receiver perform
-30
1010-30
Conventional Receiver, Red
Our Newly Proposed Receiver, Red
much better
Conventional Receiver, Yellow
-35 Our Newly Proposed Receiver, Yellow
1010-35 Conventional Receiver, Green
-40
Our Newly Proposed Receiver, Green
1010-40

1010-45
-45

6 8 10 12 14 16 18 20
6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00
Time of Day, hours

73
 Visible light communications (VLC)
for VANET
z W
Transmitter
Red
S N
ϕ
E
Yellow
Green Dd
θ
y ψ Transceiver 1 Transceiver 2
zt

yr zr
x
xt Δx = x r − x t xr

• Advantages: • Disadvantages:
– No radio interference – LOS, Short-distance transmission
– No radio spectrum required – A lot of optical noises in outdoor,
i.e., background light
– Utilization of existing infrastructure
– Safe – Changes in the relative position
between transmitter and receiver 74
– Low cost
 Receiver structure for VLC
• Background light is the major source of noise in VLC receiver
• A conventional receiver may have optical filter with relatively large
bandwidth to receive the 3 traffic light colours
– Collect higher noise level as well
• An improved structure is to have separate optical filters for green light
and red and yellow lights
– Each filter collect lesser noise

Optical Bandpass Filter

Δ λ G = λ G 2 − λG 1

Hemispherical lens
Optical Bandpass Filter
Optical Bandpass Filter
Δλ RYG = λ RYG 2 − λ RYG 1 Δλ RY = λ RY 2 − λ RY 1
RCVG RCVRY
Hemispherical lens
Silicon p-i-n photodiode
Silicon p-i-n photodiode
ARY
A Demodulator Output Selective
Preamplifier
Combining Demodulator Output
Preamplifier
Circuit
AG
Preamplifier
Conventional Selective combining
75
 L3 handoff time improvement
Two of the main L3 handoff delays:
• detection of a subnet change
• address acquisition time via DHCP

An example of L3 handoff delay improvement is to reduce the DHCP

acquisition time as in the work by Andrea G Forte, et al, ACM Proceeding on
Wireless Internet, 2006

Source: Andrea G Forte, et al,

ACM Proceeding on Wireless
Internet, 2006 76